Friday, October 12, 2012

2013 African Cup of Nations qualifying simulations

10000 simulations using the Elo Ratings.

Chances of qualification:

98.03% - Congo DR
97.75% - Mali
95.32% - Nigeria
95.20% - Ghana
93.63% - Tunisia
93.00% - Côte d'Ivoire
91.32% - Guinea

86.13% - Sudan
79.30% - Zambia
77.69% - Algeria
70.17% - Zimbabwe
69.45% - Cape Verde Islands
67.15% - Mozambique
60.09% - Central African Republic
58.88% - Gabon
41.12% - Togo
39.91% - Burkina Faso
32.85% - Morocco
30.55% - Cameroon
29.83% - Angola
22.31% - Libya
20.70% - Uganda
13.87% - Ethiopia
8.68% - Niger
7.00% - Senegal
6.37% - Sierra Leone
4.80% - Malawi
4.68% - Liberia
2.25% - Botswana
1.97% - Equatorial Guinea

Most likely composition of pots, taking into account the CAF ranking. The draw will take place on 24 October.

Pot 1: South Africa, Zambia, Ghana, Côte d'Ivoire
Pot 2: Mali, Tunisia, Gabon, Nigeria
Pot 3: Sudan, Algeria, Guinea, Mozambique
Pot 4: Congo DR, Zimbabwe, Cape Verde Islands, Central African Republic

Break-down by team:

Algeria: Pot 2 - 11.86%, Pot 3 - 65.83%, Eliminated - 22.31%
Angola: Pot 1 - 0.12%, Pot 2 - 29.71%, Eliminated - 70.17%
Botswana: Pot 3 - 1.73%, Pot 4 - 0.52%, Eliminated - 97.75%
Burkina Faso: Pot 2 - 0.41%, Pot 3 - 33.16%, Pot 4 - 6.34%, Eliminated - 60.09%
Cameroon: Pot 1 - 0.04%, Pot 2 - 30.51%, Eliminated - 69.45%
Cape Verde Islands: Pot 3 - 2.21%, Pot 4 - 67.24%, Eliminated - 30.55%
Central African Republic: Pot 3 - 13.19%, Pot 4 - 46.90%, Eliminated - 39.91%
Congo DR: Pot 4 - 98.03%, Eliminated - 1.97%
Côte d'Ivoire: Pot 1 - 93.00%, Eliminated - 7.00%
Equatorial Guinea: Pot 2 - 0.07%, Pot 3 - 1.90%, Eliminated - 98.03%
Ethiopia: Pot 4 - 13.87%, Eliminated - 86.13%
Gabon: Pot 1 - 0.04%, Pot 2 - 55.68%, Pot 3 - 3.16%, Eliminated - 41.12%
Ghana: Pot 1 - 95.20%, Eliminated - 4.80%
Guinea: Pot 2 - 4.51%, Pot 3 - 84.89%, Pot 4 - 1.92%, Eliminated - 8.68%
Liberia: Pot 4 - 4.68%, Eliminated - 95.32%
Libya: Pot 2 - 0.23%, Pot 3 - 18.06%, Pot 4 - 4.02%, Eliminated - 77.69%
Malawi: Pot 2 - 0.03%, Pot 3 - 4.00%, Pot 4 - 0.77%, Eliminated - 95.20%
Mali: Pot 1 - 29.62%, Pot 2 - 68.13%, Eliminated - 2.25%
Morocco: Pot 2 - 0.25%, Pot 3 - 23.58%, Pot 4 - 9.02%, Eliminated - 67.15%
Mozambique: Pot 2 - 0.08%, Pot 3 - 41.56%, Pot 4 - 25.51%, Eliminated - 32.85%
Niger: Pot 2 - 0.04%, Pot 3 - 6.90%, Pot 4 - 1.74%, Eliminated - 91.32%
Nigeria: Pot 2 - 73.19%, Pot 3 - 22.13%, Eliminated - 4.68%
Senegal: Pot 2 - 0.39%, Pot 3 - 5.65%, Pot 4 - 0.96%, Eliminated - 93.00%
Sierra Leone: Pot 3 - 0.26%, Pot 4 - 6.11%, Eliminated - 93.63%
Sudan: Pot 2 - 33.96%, Pot 3 - 52.17%, Eliminated - 13.87%
Togo: Pot 3 - 19.51%, Pot 4 - 21.61%, Eliminated - 58.88%
Tunisia: Pot 1 - 2.68%, Pot 2 - 90.95%, Eliminated - 6.37%
Uganda: Pot 3 - 0.11%, Pot 4 - 20.59%, Eliminated - 79.30%
Zambia: Pot 1 - 79.30%, Eliminated - 20.70%
Zimbabwe: Pot 4 - 70.17%, Eliminated - 29.83%

About me:

Christian, husband, father x 3, programmer, Romanian. Started the blog in March 2007. Quit in April 2018. You can find me on LinkedIn.

15 comments:

  1. So far:
    91.32% - Guinea OUT
    86.13% - Sudan OUT
    70.17% - Zimbabwe OUT
    67.15% - Mozambique OUT
    58.88% - Gabon OUT

    5 out of 12 were wrong (Mali, Zambia, Ghana, Cote d'Ivoire, Nigeria, Tunisia, Cape Verde)

    Pretty poor return so far...

    ReplyDelete
  2. The one flaw in ELO ratings is the 100 points for a team playing at home. Essentially suggesting home is exactly twice as likely to win as away.

    In Africa away wins are pretty rare, stats show the home/away bias is more like 75%/25%. This would require approximately a 250 point bonus for being at home using the ELO system.

    ReplyDelete
  3. Ends with 9 out of 15 predictions correct... 60%... still pretty low. And Zambia was a penalty kick away from missing out...

    Niger is the major shocker.

    ReplyDelete
    Replies
    1. That's why football matches are played - the results can be surprising some times.

      Delete
  4. The CAF World Cup qualifying groups contain the following number of Nations Cup qualified teams:

    A:3
    B:2
    C:2
    D:2
    E:1
    F:1
    G:0
    H:2
    I:2
    J:1

    With Algeria's win over Libya, I wonder whatthe FIFA/ELO disparity will be like now, they're 24th and get 105pts in FIFA for that win.

    ReplyDelete
  5. Lorric,

    Algeria has climbed to 67th spot in ELO. In the FIFA-ranking they total at 907 FIFA points for November. Looks like a spot just in or just out of the top-20 for Algeria after tomorrow's matchday.

    Still quite a disparity in classification for Algeria between ELO and FIFA though.

    ReplyDelete
  6. By the way,

    your distribution of Nations Cup qualifiers over WC qualifying groups is flawed. Group A and E both have two ACN qualifiers. Burkina Faso has qualified instead of CAR. It was a very late goal, but still... :D

    ReplyDelete
    Replies
    1. But do you know why there is such a disparity? It's been like that for as long as I can remember, about 40 places higher in FIFA for Algeria. I'd really like to know what the cause is. Other teams come and go, but Algeria are always there in Edgar's FIFA/ELO comparison posts near the head of the lists, it never changes.

      Oops. I think I wanted Central Africa to qualify too much. That late goal is just brutal with a capital B. Worse than losing on penalties surely. Never qualified before, dream crushed at the very end.

      Delete
  7. Lorric,

    It is true, from 1980 to 2000 Algeria was a regular inhabitant of the range of 30 - 40 in the ELO-ranking with a highest position of 21.

    Looking at the period starting from 2000 (Algeria was 69th then) they were never higher than 55th (!) in ELO and that was in January 2010 when they won against Ivory Coast in the quarter final of the CAN 2010. After that they lost the semi-final against Egypt and 3rd place match against Nigeria after which they were back at 71.

    In that same ACN-tournament they lost their first group match very surprisingly against Malawi which cost them 68 ELO-points and 25 ranking places.
    Another big surprise loss was a 0:3 WC-qualifier against Gabon in September 2004 which cost them 56 points and 18 places.

    In that whole decade they never had big surprise wins (in terms of ELO-points) and with the few big surprise losses I mentioned they are now still trying to get back to the top 50 in ELO.

    In the FIFA-ranking they are on the rise from July 2009 (they were 66th in June 2009). The ACN 2010 cost them next to nothing in terms of ranking position and their rise is further enhanced by good results this last year in CAF and WC qualifying. Their first year average for instance has climbed some 300 FIFA points between September 2011 and now. In that period they have climbed from 94th spot to 67th in ELO and from 46th to say 20th in FIFA.
    So the trend (and even the number of places they climbed) is the same in both rankings this last year: only in FIFA they started at a much higher position.

    Hope this sheds some light on the where-abouts of Algeria in ranking land.

    ReplyDelete
    Replies
    1. I am Algerian. Algeria has the exception to have margin wins even against the weakest teams. They often have a one-goal win against small teams. As the Elo rating counts for the goal margin unlike the FIFA ranking, this doesn't bring Algeria enough points to climb in the ranking table. Another issue, Algeria had a terrible year in 2010: out of 16 games (9 losses, 4 draws, 3 wins). The Elo rating now starts for Algeria from the game against Egypt on November 14th, 2009 but Fifa ranking counts for games a year earlier (October 2008) where Algeria has made excellent performance in the WC qualifiers and friendlies they played. They were undefeated for a year and a half year before loosing against Egypt on November 14, 2009.

      So I think that the disparity for Algeria comes from the Fifa ranking counting games for 4 years while Elo considers 3 years only (30 games rule) and from a bad 2010-year for Algeria which doesn't affect the Fifa ranking that much as it does for the Elo.

      Delete
    2. Hi Brahim and Lorric,

      I've taken another in-depth look into this matter, because I thought 'it couldn't be the few shock losses only that cause this rather enormous disparity'.
      For that purpose I have made an excel-graph with the positions in ELO and FIFA against time (it's a pity I can't show the graph here, it's very insightfull). You can see that the current difference of plus minus 40 places between FIFA and ELO has his main origin in only one game !

      And that's the match Algeria-Senegal 3:2 (a WC qualifier played on Sep 5 2008). Algeria climbed 3 places in ELO because of that win, because the ELO-points difference between them at that time was very small (Algeria 1519 (including 100 pts home-advantage) and Senegal 1552).
      In FIFA Algeria climbed 20 places because Senegal was a high-positioned opponent at the time (43) and the win delivered Algeria nearly 1000 match points.

      In the period between June 2008 and October 2008 the ELO-position of Algeria climbed from 100 to 93 (+7 places) and their FIFA-position rose from 103 to 56 (+47 places). After that the difference in position fluctuated between 20 and 60 positions but in average it stayed around the current 40 positions.

      Btw Brahim, ELO takes all matches into account, not only the last 30. The ELO-rating tends to become stable after 30 matches are played (stable with regard to the initial rating (which is nothing more than a scientific guess) that a new country gets assigned when they are incorporated in the ELO-ranking).

      Delete
    3. Ah, well done and thanks.

      So that's it. I was looking for some sort of pattern, and no wonder I couldn't find one, with it being that single one off fixture. I'm really surprised that a single fixture could cause all that. And also with it being so long ago, I don't get that, that fixture isn't even on the FIFA ranking anymore now, so why the difference? I don't understand that, shouldn't FIFA have gone back down to the same position as ELO if that match made the difference?

      Delete
    4. Well, that fixture (and maybe a second, slightly less striking, fixture in that period against Gambia, which I forgot to mention and which also was of much more influence in FIFA than in ELO) may be excluded from the FIFA-calculation by now, but it just caused the difference at that time.

      After that, both rankings in general follow the same trend until now but the difference was made once and it could well be that this difference therefore remains in the same order of magnitude for quite some time to come. The graph anyway shows no clear trend that the positions in both rankings are converging at the moment. As I said, this difference remains since 2008 on average at around 40 positions.

      I think it takes a 'reverse' situation to considerably diminish the difference: a match with a lot of ELO-points and only a limited amount of FIFA match points in it.

      Delete
    5. I guess we don't really know what has caused it then. But thanks for looking into it.

      Delete
    6. Ed, thanks for the analysis!

      Delete